RSSI variance

[Alex Cheng]

Hi !

According to application requirement, we need to minimize the RSSI variance of the BLE beacon as small as possible. Our RF team has tried to minimize the 37/38/39 channel gains within 1.3 dbm by turning the antenna design. But in our field trial (the distance between the beacon and BLE scanner is 7m in open space), we found that the RSSI variance is larger than 1.3dbm (the variance is around 4 dbm and channel 39 always has the lowest RSSI). Is it caused by voltage drop when the system is doing the TX? May I know what the root cause is ? Any suggestion ? Antenna gain-2016-02-16_141412.jpg Field trial -2016-02-16_141829.jpg

[run_ar]

Which chip are you testing with? For the nRF51822 the RSSI accuracy is +/- 6dB within the valid range of -50 dBm to -80 dBm.

[Alex Cheng]

The chip we used is nRF51822. You mean no matter the Antenna we used the RSSI accuracy is +/- 6dB within the valid range of -50 dBm to -80. May I know the root cause ?

[run_ar]

Actually this is the chip to chip production variance + variance over input power + margins. The actual on chip variance from measurement to measurement on the same channel would be more accurate, probably something closer to +/- 1-2 db. In addition there will be some variance from channel to channel due to some non-linearity. You will also get some variance because of the output on the transmitter radio (variance in radio + filter + antenna) over the frequency band. And the same variance over the input on the receiving radio. In addition you will probably have some reflections that will vary based on frequency.

[Alex Cheng]

Could you please detail " In addition there will be some variance from channel to channel due to some non-linearity." ? We do found that some channel has poor performance.

[run_ar]

The environment in which you measure this would also affect the result as you will have reflections that can vary over frequency. Note that I'm not able to see the files you attached in your questions, so I'm not sure how your measurements look.